Tectonic segmentation of the North Andean margin: impact of the

ELSEVIER Earth and Planetary Science Letters 168 (1999) 255–270

Tectonic segmentation of the North Andean margin: impact of theCarnegie Ridge collision

M.-A. Gutscher a,Ł, J. Malavieille a, S. Lallemand a, J.-Y. Collot b

a Laboratoire de Geophysique et Tectonique, UMR 5573, Universite Montpellier II, Place E. Bataillon,F-34095 Montpellier, Cedex 5, France

b IRD, Geosciences Azur, Villefranche-sur-Mer, France

Received 17 July 1998; accepted 2 March 1999

Abstract

The North Andean convergent margin is a region of intense crustal deformation, with six great subduction earthquakesMw ½ 7:8 this century. The regional pattern of seismicity and volcanism shows a high degree of segmentation along strikeof the Andes. Segments of steep slab subduction alternate with aseismic regions and segments of flat slab subduction. Thissegmentation is related to heterogeneity on the subducting Nazca Plate. In particular, the influence of the Carnegie Ridgecollision is investigated. Four distinct seismotectonic regions can be distinguished: Region 1 – from 6ºN to 2.5ºN withsteep ESE-dipping subduction and a narrow volcanic arc; Region 2 – from 2.5ºN to 1ºS showing an intermediate-depthseismic gap and a broad volcanic arc; Region 3 – from 1ºS to 2ºS with steep NE-dipping subduction, and a narrowvolcanic arc; Region 4 – south of 2ºS with flat subduction and no modern volcanic arc. The Carnegie Ridge has beencolliding with the margin since at least 2 Ma based on examination of the basement uplift signal along trench-paralleltransects. The subducted prolongation of Carnegie Ridge may extend up to 500 km from the trench as suggested bythe seismic gap and the perturbed, broad volcanic arc. These findings conflict with previous tectonic models suggestingthat the Carnegie Ridge entered the trench at 1 Ma. Furthermore, the anomalous geochemical (adakitic) signature of thevolcanoes in the broad Ecuador volcanic arc and the seismicity pattern are proposed to be caused by lithospheric tearsseparating the buoyant, shallowly subducting Carnegie Ridge from segments of steep subduction in Regions 1 and 3. Itis further suggested that Carnegie Ridge supports a local ‘flat slab’ segment similar to that observed in Peru. The impactof the Carnegie Ridge collision on the upper plate causes transpressional deformation, extending inboard to beyond thevolcanic arc with a modern level of seismicity comparable to the San Andreas fault system. The pattern of instrumentaland historical seismicity indicates (1) great earthquakes on the northern and southern flanks of the colliding ridge, (2) aslight reduction in observed seismicity at the trench–ridge intersection, (3) increased stress far into the continent, and (4) aNNE displacement of the N. Andes block, to be further effects of the collision. 1999 Elsevier Science B.V. All rightsreserved.

Keywords: Northern Andes; plate collision; tectonics; seismicity; segmentation; earthquakes

Ł Corresponding author. Fax: C33-467-523908; E-mail: [email protected]

0012-821X/99/$ – see front matter 1999 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 9 9 ) 0 0 0 6 0 - 6

256 M.-A. Gutscher et al. / Earth and Planetary Science Letters 168 (1999) 255–270

1. Introduction and tectonic setting

The North Andean margin is a region of intensecrustal deformation, in particular where CarnegieRidge is subducting beneath Ecuador (Fig. 1). Thissection of the subduction zone has produced sixgreat (defined as Mw > 7:75) earthquakes this cen-tury. The largest in 1906 (Mw D 8:8) had an esti-mated rupture zone of ca. 500 km length, partially

Panama

Rift

Malpelo R.

Malp.Rift

Grijalva F

Z

Yaqu

ina

Gra

ben

NorthAndesBlock

Alvarado R.

Sarmiento R.

Cocos R

idge

2500m

Coa

stal

R.

Coi

ba R

.

Carnegie RidgeGalapagos

Islands

Active Volcano

Explanation

Mangrove Swamp

Drainage Basin w.Outflow Direction

Subduction Zone

Thrust Fault

Oceanic Plateau(2500 m contour)

Spreading Center(active)

Spreading Center(inactive)

Coastal Relief200 m, 400 m

Transform

Strike Slip Fault

Figures 2,3

Inferred Fault orSubducted Structure

3 MagneticAnomaly

6˚S

8˚S

6˚N

8˚N

10˚N

0˚

4˚N

2˚N

2˚S

4˚S

7

89

10

3

3

2

2

2

2

5B

5A

5A

5B2

75˚W85˚W95˚W 90˚W 80˚W

Farallon Ocean Crust

4

2

3

3

4

4

5

5

Pan

ama

FZ

CocosPlate

Plate Motions

NazcaPlate

7.2

cm/a

9.1

cm/a

Global ModelNUVEL-1(De Mets et al., 1990)

GPS(Kellogg and Vega, 1995)

BrazilianShield

7.0 cm/a

7.0 cm/a

Caribbean Plate

6-7cm/a 6-7

cm/a

GG

DGM

7.0 cm/a

5.0 cm/a

Fig. 1. Tectonic setting of the study area showing major faults, relative plate motions according to GPS data [7] and the NUVEL-1 globalkinematic model [8], magnetic anomalies [13] and active volcanoes [50]. Here and in Fig. 4, the locations of the 1906 (Mw D 8:8, verylarge open circle) and from south to north, the 1953, 1901, 1942, 1958 and 1979 (M ½ 7:8, large open circles) earthquakes are shown.GG D Gulf of Guayaquil; DGM D Dolores–Guayaquil Megashear.

reactivated in three subsequent events from south tonorth, in 1942, 1958 and 1979 [1,2]. This region,frequently cited when discussing models of segmentrupture, earthquake propagation and recurrence in-tervals [1–4], was also the site of the large Mw D 7:1earthquake, August 4, 1998. The region of great-est seismicity coincides with the subducted northernflank of Carnegie Ridge. Unlike other subductionzones, in the North Andes instrumental and histori-

M.-A. Gutscher et al. / Earth and Planetary Science Letters 168 (1999) 255–270 257

cal seismicity M ½ 7 extend hundreds of kilometersinland and beyond the volcanic arc. The current de-bate on seismic coupling in subduction zones [5,6]and its relation to subducting bathymetric highs (as-perities) bears strongly on the entire area and on theassessment of its seismic risk.

Recent advances in instrumentation and computertechnology and availability of satellite-derived datapresent a unique opportunity to combine indepen-dent, high resolution, digital data sets from a va-riety of fields (seismology, volcanology, morphol-ogy=topography, satellite altimetry, GPS–geodeticstudies, structural geology) for application to geo-dynamic problems. The principal objective of thisstudy is to re-examine the seismicity and tectonicsin this region of intense crustal deformation in viewof these newly available data. By studying the spa-tial distribution of hypocenters and examining focalmechanisms we aim to identify the principal faultplanes, identify areas prone to large earthquakes andscrutinize apparent seismic gaps. Correlations be-tween structural heterogeneity of the oceanic NazcaPlate and uplift, deformation and volcanism in theoverriding South American Plate should improveour understanding of these processes and of theoverall lithospheric structure of an oceanic plateau–subduction zone collision.

Along the North Andean margin, the Nazca Plateis subducting eastwards beneath South America ata rate of 5–7 cm=a [7,8] (Fig. 1). Simultane-ously, the North Andean block is being displacedto the northeast at a rate of ca. 1 cm=a along theDolores–Guayaquil Megashear accounting for thediscrepancy between the global model (Nazca mo-tion relative to stable South America) [2] and theGPS measurements (Nazca motion relative to theNorth Andes block [7,9] (Fig. 1). Thus, the ENE-ori-ented Carnegie Ridge is sweeping southwards alongthe Ecuador margin. Recent coastal uplift facingCarnegie Ridge is indicated by Pliocene marine ter-races (Tablazos) exposed at 200–300 m elevations[10,11]. To the north, the morphology, widespreadmangrove swamps and neotectonics of the SouthColombian coastal block, suggest active subsidence[11].

The most widely accepted tectonic model for theformation of the oceanic crust of the E. PanamaBasin suggests that a major plate reorganization took

place ca. 25 Ma, breaking the Farallon Plate intothe Cocos and Nazca plates to the south and theJuan de Fuca Plate farther to the north [12]. Dueto differential stresses on the northeastward-subduct-ing Cocos Plate and the eastward-subducting NazcaPlate, spreading was initiated around 23 Ma be-tween the two plates in the vicinity of the Galapagoshotspot and later evolved into the current-day Gala-pagos Rift with N–S spreading [13,14]. The Grijalvascarp, an old N60ºE fracture zone in the FarallonPlate, is interpreted to represent the southern half ofthe scar where the Nazca Plate tore off. Carnegieand Cocos ridges are mirror-image hotspot tracesformed by the northeastward motion of the CocosPlate and the eastward motion of the Nazca Plateover the Galapagos hotspot [13–16]. According tothis model both Cocos and Carnegie ridges arrived attheir respective trenches 1 Ma [13].

Most models suggest that Malpelo Ridge is aformer continuation of Cocos Ridge separated bydextral motion along the N–S-trending Panama FZ[13,14]. Malpelo and Carnegie ridges separated fromone another during a period of rifting and seafloorspreading from ca. 17 to 8 Ma [13,15,17]. The ex-pressions of the extinct Malpelo Rift and adjacentrift segments as well as the transform faults separat-ing them are clearly visible in the morphology of theseafloor (Fig. 2).

2. Seismicity database and observations

Earthquake data in the study area were obtainedfrom Engdahl et al.’s recent global relocation effort[18], improving hypocenter locations from the Inter-national Seismological Centre (ISC) Catalog. Thus,for this study area 1230 events Mb > 4:0 fromJanuary 1964 to December 1995 were available. Cat-alogs of lesser quality but (1) covering a greater timespan (the South American SISRA Catalog [19] ca.800 events Mb > 4:0 from 1900 to 1972), or (2)comprising more events (the USGS PDE – Prelim-inary Determination of Earthquakes – Catalog; ca.1800 events Mb > 4:0 from 1973 to 1997), were alsoexamined and revealed the same overall distributionpatterns, though with inferior spatial resolution andwell known ‘artifacts’ such as large concentrationsof events at depths of 33 km or 15 km.



The EHB database [18] is shown sorted by mag-nitude and hypocenter depth and was used to cal-culate depth contours to the Wadati–Benioff zone(Fig. 2). Four different regions, with distinct pat-terns of seismicity and volcanic activity [15,20] areobserved from north to south. Representative litho-spheric cross-sections based on the seismicity, vol-canism and topography databases are presented toillustrate these main differences (Fig. 2).

Region 1, Section A–A0. The northernmost region(in west-central Colombia) shows a steep ESE-dip-ping Wadati–Benioff zone down to a depth of about200 km in good agreement with earlier seismolog-ical [15] and recent tomographic results [21]. Thevolcanic arc is narrow and located above a slab atca. 160 km depth. Malpelo Ridge and the activePanama FZ lie west. Lithospheric ages (<5 Ma) andthicknesses (<20 km) are least to the west sincenearby at the Panama Rift, new oceanic lithosphereis currently forming (Fig. 1).

Region 2, Section B–B0. Between 2.5ºN and 1ºSthere is a nearly complete absence of intermediatedepth seismicity (only one event >90 km depth).The seismic gap is confirmed by the SISRA andPDE catalogs as well as by a local seismologicalnetwork of 54 stations operated during the Litho-scope experiment [22]. Carnegie Ridge enters thetrench across from the broad (160 km wide), adakiticvolcanic arc. The inferred subducted continuation ofCarnegie Ridge is interpreted to be a ‘flat slab’ con-figuration, but is constrained by only a single majorevent Mb D 5:8 at 120 km depth.

Region 3, Section C–C0. Between 1ºS and 2ºSthere is cluster of intermediate-depth seismicitydefining a steep NE-dipping slab. South of CarnegieRidge the transition from young to older oceaniclithosphere (Grijalva FZ) is crossed. The narrower1.2º sampling width and ENE alignment clearly im-

Fig. 2. Shaded hill relief and seismicity of the study area. Bathymetry and topography from Smith and Sandwell’s TOPEX database[51] with active volcanoes (red triangles) [50]. Seismicity (1964–1995), 1230 events Mb > 4:0, from Engdahl et al.’s global hypocenterrelocation [18] scaled by depth and magnitude, omitting upper plate seismicity (<70 km depth >200 km east of the trench) in map.Oceanic plateaus defined by 2500 m contour. Location and sampling boxes of seismological sections indicated. Depth contours tothe Wadati–Benioff zone indicated as dotted lines. Seismotectonic Regions 1–4 (see text) also shown. Section A–A0, Region 1, steepESE-dipping subduction, narrow volcanic arc. Section B–B0, Region 2, intermediate-depth seismic gap, subduction of Carnegie Ridgewith inferred flat slab shown, broad volcanic arc is spread out over 150 km. Section C–C0 , Region 3, narrow volcanic arc, steepNE-dipping slab. Section D–D0, Region 4, Peru flat slab segment, no volcanic arc. Section F–F0, Andes-parallel profile illustrating theintermediate depth seismic gap and inferred Carnegie ‘flat slab’ in Region 2. Note the tear north of the steep NE-dipping slab in Region 3.

age a steep-dipping slab segment with its narrowoverlying arc.

Region 4, Section D–D0. South of 2ºS the Peru‘flat slab’ segment begins, with a volcanic gap all theway to 15ºS in southern Peru [23–26]. Nearly hor-izontal subduction is observed, despite the greaterage and presumably greater density and thicknessof Farallon crust. Recent work suggests that this1500 km long ‘flat slab’ is supported by two buoy-ant oceanic plateaus, the completely subducted IncaPlateau in N. Peru and the currently subductingNazca Ridge in S. Peru [27]. Since no interveningwedge of asthenospheric material is present betweenthe downgoing Nazca Plate and the overriding SouthAmerican Plate, partial melting is inhibited and noarc develops. Shallow events (<40 km) in the E.Andes (between km 700 and 780) appear to repre-sent intracontinental ‘subduction’ where the Brazil-ian shield underthrusts the Andes at a rate of ca. 1.5mm=a, producing crustal-scale thrust slices [28].

While the northern two sections are orientednearly parallel to the relative convergence directionbetween the North Andes block and the Nazca Plate[7,9], the southern sections are oriented ENE–WSWin order to better sample the particular region. Notethat the ‘seismological thickness’ of the subductingoceanic lithosphere in all profiles is ca. 30 km, aspointed out by earlier investigators [23]. This cor-responds roughly to the effective elastic thickness.Te/ of the lithosphere as related to its thermal agethrough well known cooling relationships. For a 40Ma oceanic lithosphere an effective elastic thicknessof ca. 25 km is predicted for a wet olivine mantlerheology, while a 100 Ma lithosphere is predicted tohave Te D 35 km [29].

Finally, the Andes parallel section F–F0 (Fig. 2)also clearly illustrates the principal differences be-tween the four regions. Furthermore, it implies that


Table 1Listing of events and fault plane solutions obtained from the Harvard CMT catalog [30]

Event No. Date Time Lat. Long. Depth Mb Mw Fault plane 1 Fault plane 2 Notes(d=m=y) GMT (º) (º) (km) strike=dip=slip strike=dip=slip mech.=loc.

(º) (º)

1 30=07=80 06:56 5.18 �82.70 18.0 5.8 6.3 4=72=�171 272=81=�19 SS PanFZ2 26=11=83 09:22 3.96 �82.52 32.7 5.4 5.6 174=83=173 265=83=8 SS PanFZ3 12=12=88 13:32 5.31 �82.49 6.2 5.3 5.6 181=90=180 271=90=0 SS PanFZ4 04=02=89 19:24 5.81 �82.63 10.9 5.8 6.3 93=76=10 0=80=166 SS PanFZ5 20=07=90 04:05 4.70 �82.63 1.0 5.0 5.1 176=90=180 266=90=0 SS PanFZ6 15=08=77 20:23 2.66 �84.51 17.7 5.2 5.0 251=45=�90 71=45=�90 N PanRift7 18=09=92 10:50 3.39 �82.96 22.2 5.1 5.4 254=45=�90 74=45=�90 N PanRift8 08=05=77 16:45 �1.25 �81.08 29.6 5.1 5.5 336=16=63 184=76=97 T subd9 01=03=79 14:33 0.61 �80.06 9.6 5.6 5.7 23=24=116 175=69=79 T subd10 12=12=79 08:00 1.57 �79.36 26.6 6.4 7.9 30=16=118 181=76=83 T subd11 26=01=80 15:27 2.27 �79.57 26.1 5.0 5.6 12=22=76 207=69=95 T subd12 06=05=81 21:36 �1.92 �80.94 15.1 6.0 6.4 339=17=81 168=73=93 T subd13 10=06=85 03:23 3.00 �78.51 29.2 5.6 5.5 32=19=125 176=74=79 T subd14 25=08=90 11:43 5.75 �77.53 28.4 5.1 5.3 350=36=79 183=54=98 T subd15 02=09=90 04:27 �0.16 �80.25 14.6 5.8 6.2 22=27=115 174=65=78 T subd16 19=11=91 22:29 4.53 �77.38 25.0 6.5 7.2 13=13=95 188=77=89 T subd17 07=07=81 10:25 2.71 �79.82 17.8 5.1 5.3 16=43=�78 180=48=�101 N Yaq. B18 12=02=89 20:03 2.77 �79.76 4.9 5.0 5.1 340=46=�116 195=50=�66 N Yaq. B19 09=09=89 01:40 2.46 �79.73 14.8 6.0 5.7 1=29=�123 218=66=�73 N Yaq. B20 26=11=94 04:48 2.87 �79.43 15.9 5.1 5.3 357=37=�114 206=57=�73 N Yaq. B21 13=11=95 12:36 2.92 �79.39 17.6 5.3 5.3 208=45=�90 28=45=�90 N Yaq. B22 02=01=81 07:37 2.12 �79.16 14.2 5.7 5.9 26=42=�128 252=58=�61 N tear?23 07=01=81 07:01 1.98 �79.26 42.5 5.7 5.9 32=33=�114 240=60=�75 N tear?24 18=08=80 15:08 �2.02 �80.04 55.5 5.5 5.9 19=27=36 256=75=�112 S tear?25 27=06=81 21:54 �3.06 �80.28 53.1 5.1 5.3 298=30=�20 45=80=�119 S tear?26 10=02=90 17:12 �3.20 �80.73 54.7 5.6 5.5 138=58=167 235=79=32 S tear?27 19=05=83 19:07 0.14 �77.10 24.1 5.6 5.2 60=49=129 189=54=54 ST EAnd28 06=03=87 01:54 �0.01 �77.66 12.2 6.1 6.4 198=20=118 348=73=81 T EAnd29 06=03=87 04:10 0.04 �77.78 19.5 6.5 7.1 195=27=98 7=64=86 T EAnd30 06=03=87 08:14 0.07 �77.85 8.6 5.4 6.0 226=40=�166 125=81=�51 SS EAnd31 11=08=90 02:59 �0.18 �78.53 20.2 5.0 5.1 323=45=53 190=55=122 ST Quit32 22=09=87 16:21 �1.06 �78.01 19 5.8 6.0 197=42=129 330=59=61 ST IAD33 26=12=92 14:57 �1.01 �78.02 6.8 5.8 5.4 200=46=166 300=80=45 ST IAD34 17=01=84 16:19 �3.93 �81.40 35.4 5.9 5.5 303=45=78 139=47=101 T Amot35 29=07=90 15:29 �4.94 �81.00 41.5 5.4 5.2 10=17=58 223=76=99 T Amot36 29=03=91 20:13 �3.98 �80.92 38.2 5.2 5.6 55=27=151 171=77=66 T Amot37 11=04=82 12:21 �2.84 �78.63 15 5.4 5.1 39=26=56 256=68=106 T ASA38 03=10=95 01:51 �2.83 �77.85 43.5 6.5 7.0 234=39=120 18=57=68 T SbAnd39 03=10=95 12:45 �2.81 �77.82 15.2 6.0 6.4 243=45=157 349=74=47 ST SbAnd40 08=10=95 10:27 �2.69 �77.89 33.6 5.3 5.4 240=80=�170 148=80=�10 SS SbAnd41 31=03=83 13:13 2.41 �76.63 29.3 5.4 5.6 26=76=175 117=85=14 SS Col42 06=06=94 20:47 2.79 �75.97 5.4 6.4 6.8 206=76=170 299=80=14 SS Col43 20=04=80 11:26 �2.58 �78.52 101.1 5.1 5.2 151=35=�90 331=55=�93 dp.EcPer44 08=10=80 22:01 �1.40 �77.63 199.5 5.5 6.0 115=16=�117 323=76=�82 dp.EcPer45 03=11=81 07:02 �1.87 �78.36 141.8 5.5 5.9 133=28=�99 323=62=�85 dp.EcPer46 18=11=82 14:57 �1.72 �76.65 193.2 6.0 6.5 163=29=�51 301=68=�109 dp.EcPer47 12=04=83 12:07 �4.86 �78.10 90.3 6.5 7.0 339=35=�86 153=55=�93 dp.EcPer48 11=09=93 06:14 �4.71 �76.27 120.2 5.6 5.7 154=44=�109 359=49=�73 dp.EcPer49 29=05=79 12:59 5.24 �75.71 116.9 4.9 5.0 174=44=�100 8=47=�80 dp.Col50 23=11=79 23:40 4.78 �76.17 107.8 6.4 7.2 137=41=�163 35=79=�51 dp.Col51 25=06=80 12:05 4.48 �75.77 165.4 5.7 6.0 231=74=14 137=76=164 dp.Col52 15=08=92 19:02 5.11 �75.60 116.8 5.7 5.9 228=22=�71 28=69=�97 dp.Col

Hypocenter locations from Engdahl et al. [18].


the continuation of Carnegie Ridge supports a lo-cal ‘flat slab segment’ at ca. 100–140 km depth,bounded to the south (and possibly to the north) by alithospheric tear.

A total of 189 fault plane solutions for the studyarea were available from the Harvard centroid mo-ment tensor (CMT) catalog [30] for the period Jan-uary 1976–December 1997. A representative subsetof 52 of these events is listed in Table 1 and shown inFig. 3 using the more reliable EHB hypocenter loca-tions [18] to highlight areas of particular interest andmore complicated structure, not readily explained bycurrent models.

In general, the mechanisms are consistent withthe plate tectonic model of the study area. Severalprincipal types can be distinguished: (1) shallowdextral strike-slip in the active Panama FZ (events 1–5); (2) shallow normal faulting in the active PanamaRift (events 6, 7); (3) shallow underthrusting in thesubduction zone (events 8–16); (4) shallow normalfaulting in the flexural bulge east of Yaquina Graben(events 17–21); (5) shallow thrusting and dextralstrike-slip events in the upper plate (events 27–42);and (6) deeper, commonly normal faulting events inthe downgoing slab (events 43–52).

3. Neotectonics and recent vertical motions

The ‘basement uplift signal’ of Carnegie Ridgewas determined along a reference profile 100 kmwest of the trench and compared to profiles along acurved track parallel to the course of the ColombiaTrench (Fig. 4). This was done to establish how fararc-ward the ‘basement uplift signal’ can be identi-fied. Carnegie Ridge, together with the step from theGrijalva scarp, has a 400 km width and a height ofca. 2 km (Fig. 4a). The axial trench profile clearlyshows the subducting ridge where the trench floorshallows by 1–1.5 km with respect to regions northand south (Fig. 4b). After adding the 1–1.5 km sed-imentary fill at the trench north of the ridge, the full2–2.5 km amplitude basement uplift signal is ob-tained. Beneath the Coastal Range 110 km from thetrench, the regional bathymetric=topographic profileagain shows a pronounced uplift in the continuationof the Carnegie Ridge (Fig. 4c). The elevations ofthe Coastal Range are ca. 1.5 km higher than the

base level in the submarine regions at the northernend of the profile. If recent sedimentary fill is takeninto account the difference is again 2–2.5 km. Tothe south the picture is obscured by the dramaticvariations in basement level in and around the Gulfof Guayaquil. Here the 8 km deep Progreso Basinand the 10 km deep Jambeli and Esperanza basins inthe Gulf of Guayaquil are believed to have formedas pull-apart basins within the strike-slip regime ofthe Dolores–Guayaquil Megashear (DGM) [31–33].The signal of the Carnegie Ridge induced basementuplift is less clear in the Inland Valley profile 180 kmeast of the Colombia Trench (Fig. 4d). The basementis substantially higher in this region, but poorly con-strained sediment thicknesses to the north, the 10 kmdeep Jambeli Basin to the south and the overlyingManabi Basin, render interpretation uncertain.

Pliocene marine terraces of the Tablazo Forma-tion [10,11] exposed at 200–300 m elevation in theCoastal Range and the regional morphology are con-sistent with the southward migration of CarnegieRidge according to the relative plate motions ob-tained by GPS measurements [7]. The young topog-raphy (in excess of 400 m) of the Coastal Range liesdirectly across the trench from Carnegie Ridge. Tothe north, in S. Colombia, the coastal block is cur-rently subsiding as shown by vertical motions duringthe 1979 earthquake [34] and by the coastal man-grove swamps [13]. The margin here is also markedby large forearc basins [35,36] indicating subsidenceof the former continental shelf. This pattern of recentuplift across from a subducting high and subsidenceto the north indicates collision since at least 8 Ma.

4. Geodynamic model

Based on four principal pieces of evidence wepropose that a prolongation of Carnegie Ridge hasbeen subducting for at least 2 Ma and most likely 8Ma to a position beneath the Andes, 400 km fromthe trench.

(1) The regional bathymetric=topographic profilesindicate a basement uplift signal comparable in sizeto Carnegie Ridge extending at least 110 km eastfrom the trench. A N–S drainage divide 180 kmeast of the trench lies along the prolongation of theCarnegie Ridge crest.


Fig. 3. Earthquake fault plane solutions from the Harvard CMT (centroid moment-tensor) [30] catalog. Solutions for shallow events(<50 km) are shaded black and deeper events (>50 km) gray, for 52 representative events listed in Table 1 using EHB [18] hypocenterlocations. Several principal types can be distinguished: (1) shallow dextral strike-slip in the Panama FZ (events 1–5); (2) shallow normalfaulting in the Panama Rift (events 6, 7); (3) shallow underthrusting in the subduction zone (events 8–16); (4) shallow normal faulting inthe flexural bulge east of Yaquina Graben (events 17–21); (5) shallow thrusting and dextral strike-slip in the upper plate (events 27–42)and deeper, commonly normal faulting events in the downgoing slab (events 43–52). Candidate tearing events in the north (events 22,23) and south (events 24, 25) are indicated. Bathymetry (1000 m contours) [51] and location of regional topographic profiles (Fig. 4) alsoshown, with inferred positions of Carnegie Ridge prolongation and lithospheric tears dashed.

(2) The coastal morphology (Coastal Range uplift,subsidence to the north in the S. Colombian coastalblock) are consistent with the southward migration of

Carnegie Ridge according to relative plate motionsand indicate collision since 8 Ma.

(3) A pronounced gap in intermediate depth (70–


-6

-4

-2

0

2

40 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

-6

-4

-2

0

2

40 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

-6

-4

-2

0

2

40 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

-6

-4

-2

0

2

40 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

Yaq. Gr. Ridge lower slope /trench fill

Malpelo RidgeCarnegie Ridge

Grijalva FZ

trench fill Grijalva FZ

pelagic seds.

forearc basin fill

Coastal Range withTablazos

Undisturbed oceanic crust

Axial trench profile

Coastal Range profile

Inland valley profile

V

X

Y

Z

V'

X'

Y'

Z'

N S

Gulf ofGuayaquil

JambeliBasin

AmotapeRange

Amotape Range w.Tablazos

Gulf ofGuayaquil

ProgresoBasin Esperanza

Basin

Manabi Basin

forearc basin fill (?)

Carnegie Ridge

Carnegie Ridgeuplift

? Carnegie Ridge ?? uplift ?

a)

b)

c)

d)

Fig. 4. Four regional topographic profiles parallel to the Colombia Trench axis (locations in Fig. 3). Basin and basement structure (forplace names see Fig. 6) adopted from earlier studies [31–33]. (a) Section V–V0, undisturbed oceanic crust profile, 110 km west of thetrench. Note Malpelo Ridge to the north and Grijalva FZ and two adjacent ridges in the Farallon Ocean Crust to the south. (b) SectionX–X0, trench axial profile. Note ca. 1–1.5 km thick trench fill north of Carnegie Ridge [35,36] and south of Grijalva FZ [31]. The full400 km ð 2 km signal of the Carnegie Ridge is visible. (c) Section Y–Y0, Coastal Range profile, 110 km east of trench. Note the 1.5 kmdifference in the base level between the Coastal Range and the submerged coastal block in S. Colombia. (d) Section Z–Z0, Inland Valleyprofile, 180 km east of trench. Note the 8–10 km thick sediments in the Gulf of Guayaquil area at the southern end of the profile.

200 km) seismicity is observed in the prolongation ofCarnegie Ridge, while normal seismicity continuesin a steep (ca. 30º) Wadati–Benioff zone both northand south of this region. Seismic gaps are commonlyobserved where an oceanic plateau or other structuralhigh (e.g. seamount chain, spreading center) collideswith the trench [3,37].

(4) The active volcanic arc is broad (spread outover 150 km) and geochemically anomalous (see

below). Therefore, some major geodynamic elementis perturbing the system at depth.

The simple model of a continuation of CarnegieRidge meets with two principal objections:

(1) It conflicts with the widely accepted view thatthe oceanic plateau terminates at the fossil YaquinaFZ and that collision with the trench began recently(1 Ma) [13,33].

(2) If Carnegie Ridge has been subducting for 8


Ma, why does arc volcanism continue in contradic-tion to the model of a ridge-induced volcanic gap[37]? And why is steep subduction [15,38] observeddirectly to the north and to the south?

The opinion that Carnegie Ridge does not con-tinue east of the Yaquina Graben is based on ge-ometric considerations, on poorly constrained mag-netic anomaly data and on the presumption that thehotspot trace began at 25 Ma [13,17]. The ‘termi-nation model’ further proposed that both CarnegieRidge and Cocos Ridge (Costa Rica) entered theirrespective trenches at 1 Ma [13]. However, geolog-ical data on uplift rates in Costa Rica, indicatingstrong uplift as early as 3 Ma [16], contradict thisinterpretation for the Cocos Ridge collision and callinto question its validity for the collision of its mirrorimage, Carnegie Ridge.

Recent studies suggest that the Caribbean Plateauis a large igneous province (LIP) formed as the plumehead of the Galapagos hotspot rose to the surface, pro-ducing immense quantities of flood basalts in the LateCretaceous ca. 90 Ma, [39,40]. Later Caribbean floodbasalt-type magmatic events occurred from 76 Ma to63 Ma [40]. Thus, if the hotspot was active in the Cre-taceous and from late Tertiary to Quaternary times(25 Ma to present), it seems likely that the plume andassociated hotspot trace existed between 60 Ma and25 Ma. But if Carnegie Ridge has a subducted contin-uation as suggested by some workers [15,31,41], whydoes arc volcanism continue?

Ecuador arc magmas have an anomalous ‘adaki-tic’ geochemical signature requiring either unusuallyhigh temperatures allowing fusion of oceanic crustat shallow depths (<120 km) or a severe contamina-tion by underplated basaltic crustal melts containinganomalously low concentrations of heavy rare earthelements [42]. However, a recent study on the geo-chemistry of Ecuador volcanoes virtually eliminatesthe possibility of crustal contamination and con-cludes that depth of melting is the predominant causeof geochemical variation [43]. It has been suggestedin similar arcs showing adakitic signatures that alithospheric tear or slab window is present allowingmelting of the edges of the oceanic crust in contactwith the hot asthenosphere below [44]. The geo-chemically anomalous melts originate at shallowerdepths and produce a broader volcanic arc than atypical subduction-related arc.

We propose a new model that combines the twoideas of a lithospheric tear and a continuation ofCarnegie Ridge (Fig. 5). We suggest that two litho-spheric tears bound the Carnegie Ridge collisionzone. Melting of the oceanic lithosphere at the edgesof the tears can thus explain the adakitic magmasobserved within the broad volcanic arc of Ecuador.The southern tear, imaged in Fig. 2, occurs along theGrijalva FZ separating the Oligocene age Faralloncrust to the southeast from the Cocos–Nazca crust(<25 Ma) to the northwest as has been suggestedby previous workers [15,20]. The older, denser Far-allon crust between 2ºS and 0.5ºS plunges steeply tothe northeast. The northern tear in the lithosphere istentatively located just north of the Carnegie Ridgealong the extension of the Malpelo Rift fossil spread-ing center. To the north of this region the Cocos–Nazca crust (possibly attached to older Farallon crustat depth) subducts steeply to the east-southeast. Be-tween these two tears we propose a flat slab segmentwith a configuration similar to that observed in Peruand central Chile. The buoyant prolongation of theCarnegie Ridge is interpreted to continue at least110 km and probably up to 500 km from the trench(Figs. 3 and 5). An analysis of density and buoyancyin subduction zones suggests that oceanic plateausmust have a basaltic crustal thickness in excess of 17km in order to resist subduction [45]. The westernportion of Carnegie Ridge beneath the GalapagosIslands reaches a maximum crustal thickness of 18km [46] and, though no geophysical data are avail-able on the crustal structure at the eastern end of theridge, it is presumably similar.

An alternative hypothesis of ‘slab detachment’also succeeds in explaining many of the seismologi-cal and volcanological observations. This hypothesisis based in part on the ridge termination model [13]and presumes that at a short distance from the trench(ca. 100 km) Carnegie Ridge ends abruptly and thatfurther to the east there is a section of denser, Eoceneage Farallon crust. The older, denser lithosphere mayhave broken off and be sinking into the mantle cre-ating a slab window above. This slab window wouldbe responsible for the intermediate-depth seismic gapand would allow fusion of the edges of the oceaniccrust producing the adakitic signature in the volcanicarc at the surface. However, we favor the two-tearmodel, because it implies a continuity of Carnegie


Peru"flat slab"

steep ESE dipping

CarnegieRidge

Ntear

Grijalva FZ

MalpeloFossil Rift

olderFarallon crust

<25 Ma oldCocos-Nazca

crusttrench axis

Stear

normal faultsat bulge

steep NE dipping

C.F.S.

Fig. 5. 3-D view of the two-tear model for the Carnegie Ridge collision featuring: a steep ESE-dipping slab beneath central Colombia;a steep NE-dipping slab from 1ºS to 2ºS; the Peru flat slab segment south of 2ºS; a northern tear along the prolongation of theMalpelo fossil spreading center; a southern tear along the Grijalva FZ; a proposed Carnegie flat slab segment (C.F.S.) supported by theprolongation of Carnegie Ridge.

Ridge with the Caribbean–Galapagos Plume traceand thus succeeds in explaining the coastal patternof subsidence to the north and uplift in the coastalrange in accordance with a southward-sweeping ofCarnegie Ridge.

The proposed lithospheric tears can be expected toproduce earthquakes with a steep E–W-striking focalplane and a steep slip vector. Such ‘tearing events’would be the only definitive validation of the model.While no such events have been reported in earlierstudies, two candidate events occurred in N. Ecuadorin January 1981, .Mb D 5:6–5:7/ with a focal planestriking N70ºE and steep 60º dip (events 22, 23 inTable 1 and Fig. 3). They occurred in the S. Colom-bian forearc at depths of 14 and 43 km in the vicin-ity of subduction zone thrust earthquakes (Fig. 3).They cannot be extensional events due to flexure ofthe oceanic lithosphere because (a) the flexural bulgeis located 100 km further west (see events 17–21 inFig. 3), and (b) the fault plane solutions are distinctlydifferent (Fig. 3). Since extension along the conjugateN30ºE focal plane seems unlikely given the WNW–ESE compressional stress field, down to the north mo-tion along the N70ºE-striking fault is more plausible.

Another supporting observation is the substantialseismicity in the flexural bulge (five normal eventsMb D 5:1–6:0 between 1981 and 1995), limited toa ca. 100 km long segment of Nazca Plate east ofthe Yaquina Graben (Fig. 3). This area correspondsto Region 1 where the oceanic lithosphere subductsmost steeply and thus flexure is the greatest. Yet seis-mic activity appears to terminate southwards near theproposed lithospheric tear. We interpret this patternand the normal mechanisms as indications that thelithosphere just north of the tear is subject to thegreatest flexural stresses causing surficial extensionin the bulge. Further south the buoyant CarnegieRidge causes the subducting slab to bend at a re-duced angle of curvature, thus lowering the flexuralstress in the bulge (Fig. 5).

There are candidate ‘tearing events’ to the south(events 24–26 in Table 1 and Fig. 3) with a NE–SW-striking focal plane (parallel to the Grijalva FZtrend) and down to the south sense of motion. Theyare located at 53–56 km depth and have sub-verticalfault planes dipping 75–80º. The substantial depth,and their position well west of the Andes indicatethat they are within the downgoing oceanic litho-


sphere. Additionally, there is a band of seismicityparalleling the Grijalva FZ and extending from theGulf of Guayaquil down to the 200 km deep clus-ter seismicity (from event 26 to 44 in Fig. 3). Thissuggests that the opening of the Gulf of Guayaquiland its 10–12 km deep basins may be a surfaceexpression of the lithospheric tear at depth.

5. Interplate coupling and impact on the upperplate

The Carnegie Ridge collision appears to have af-fected the coupling between the Nazca and SouthAmerican plates. Four great earthquakes have oc-curred on the northern flank of the collision (1906,1942, 1958 and 1979), and one has occurred alongthe southern flank (1901) (though the magnitude andlocation for the latter are uncertain). None of theseevents appears to have ruptured across the ridgeitself. This pattern is similar to the Nazca Ridgecollision in S. Peru. During the last 60 years fourgreat earthquakes Mw ½ 7:8 occurred north of NazcaRidge and two to the south, but again no shockappears to have ruptured across the ridge–trenchintersection [1,4].

The recent shallow (<70 km) seismicity inEcuador suggests a gap between 1ºN and 0.8ºS(Fig. 2). This appears to confirm the model thatthe subduction=collision of bathymetric highs lo-cally increases coupling [6] and causes short-termseismic gaps. Recurrence intervals in such zones ofincreased coupling may be as much as twice as longand thus exceed 100 years, leading to gaps in recentseismicity maps. Greater stress accumulation duringan aseismic period implies accordingly greater stressrelease during rupture [6]. Indeed the August 4, 1998.Mw D 7:1/ event at 0.5ºS occurred in an apparentlyquiet region, but which had been correctly identi-fied by Nishenko in 1991 [47] as a region with a66% probability of a large or great earthquake before1999.

On the other hand, the regions between 1º and2ºS and between 4º and 6ºS appear to be failingfrequently by moderate earthquakes (10 events and23 events, respectively, Mb > 5:0 within 100 kmfrom the trench from 1964 to 1995) and do notappear to present a major seismic risk. However, the

Gulf of Guayaquil segment of the subduction zone(from 2.2ºS to 3.8ºS) has seen no earthquakes Mb >

5:0 in the past 30 years. Either this region is largelyaseismic, due perhaps to the propagating lithospherictear here or, here too, seismic coupling may be quitehigh and possibly building up to a magnitude 7or greater event. The SISRA catalog [19] indicatesthat two very large, shallow earthquakes, M ½ 7:4occurred near 3.5ºS in 1953 (Fig. 1) and 1959 andthus tends to support the latter interpretation.

Interplate coupling at a much deeper and largerscale may also be affected by the Carnegie Ridgecollision. In the region facing Carnegie Ridge, in-creased upper plate seismicity and deformation ex-tending 500–600 km inland to beyond the volcanicarc suggest the collision to be the driving mecha-nism behind the NE movement of the North Andeanblock. This movement is distributed along a net-work of transpressional faults at the southern edgeof the N. Andes block [32,48,49] (Fig. 6). Individualfault splays visible in the morphology commonlycorrelate with bands of seismicity. Yet, some faultsinclude segments of ca. 100 km length without re-cent activity (Fig. 6). These segments can also beconsidered potentially high risk areas.

Overall, the level of seismicity is comparableto that of the San Andreas fault. For a 500 kmlong segment of the DGM, 41 Mb ½ 5:0 and 5Mb ½ 6:0 non-subduction related, shallow events(<70 km depth) occurred during the period Jan-uary 1964–December 1995. Along the 800 km longsegment of the San Andreas fault system from SanFrancisco to the Mexican border 80 Mb ½ 5:0 and9 Mb ½ 6:0 events were recorded during the same32 year period [18]. Historical intraplate seismicityin the Ecuadorian Andes is very high as well. Fourearthquakes of intensity 7–8, five of intensity 9, threeevents of intensity 10 and one of intensity 11 struckthe Andean region between 0º and 4ºS from 1541 to1906 [1,19] (Fig. 6).

Unfortunately, little is known about the slip ratesalong individual faults and even less about the sys-tem as a whole. Two studies have attempted toconstrain slip rates along these faults. The first one,along a splay of the Pallatanga fault in the centralEcuador Andes, is based on displacement of mor-phological features and arrived at a slip rate of 3–4.5mm=a [49]. The second study was performed along


Fig. 6. Detailed topography and shallow (<70 km depth) seismicity of the Ecuadorian Andes showing known (solid) and inferred(dashed) faults. Dolores–Guayaquil Megashear (DGM) shown bold. Pall. F D Pallatanga Fault. Topography from GTOPO-30 digitaldatabase. Filled circles from the EHB data set (1964–1995) [18], Mb ½ 4:0 scaled by magnitude. Filled squares (1900–1963), Mb ½ 5:0from the SISRA Catalog [19] scaled similarly. Unfilled black circles are epicenters of 13 large historical earthquakes from 1541 to 1906from the SISRA Catalog [19] scaled according to magnitude and intensity from MI D 5:7, I D 7 small circles to MI D 8:3, I D 11(largest circle – 1797).


the Rio Chingual–La Sofia fault at the Ecuador–Colombia border based on displacement of datedpyroclastic flows and arrived at a slip rate of 7 š 3mm=a [48]. However, in each case these are indi-vidual fault traces within a network of several faultsand if all are active to a comparable degree thentotal displacement could approach 2 cm=a. The mostreliable method to ascertain total strain rates wouldbe a geodetic (GPS) study from the Brazilian cratonacross all the DGM fault splays. Given the high levelof modern and historical intraplate seismicity in theEcuadorian Andes such a study is warranted andnecessary.

6. Conclusions

(1) The seismicity and active volcanism along theNorth Andean margin between 6ºN and 6ºS definefour distinct regions.– Region 1: from 6ºN to 2.5ºN, steep ESE-dipping

subduction, narrow volcanic arc.– Region 2: from 2.5ºN to 1ºS, intermediate depth

seismic gap, broad volcanic arc.– Region 3: from 1ºS to 2ºS, steep NE-dipping

subduction, narrow volcanic arc.– Region 4: south of 2ºS, Peru flat slab segment, no

volcanic arc.(2) A prolongation of Carnegie Ridge is proposed

to continue >110 km beyond the trench. Four piecesof evidence support this conclusion.

(a) Coastal morphology: uplift in Coastal Range,subsidence in S. Colombian coastal block, consis-tent with southward migration of Carnegie Ridgeaccording to GPS motions.

(b) Regional topography: a 400 km ð 2 kmbasement uplift signal at least 110 km east of thetrench.

(c) Seismicity pattern: intermediate depth seismicgap.

(d) Volcanic arc: a broad arc spread out over 150km, the anomalous geochemistry of the arc magmas.

(3) The geodynamic segmentation and anomalousgeochemistry (adakitic signature) of the arc magmasimplies one or more lithospheric tears. The favoredmodel involves two tears, at the boundaries betweenregions 1–2 and regions 2–3. The bowknot contin-uation of Carnegie Ridge is interpreted to support a

local flat slab segment with melting of the oceanicplate at the edges of the tears producing adakiticmagmas.

(4) Modern and historical intraplate seismicityis very high in the Ecuadorian Andes and in thenetwork of NE-striking transpressive faults at thesouthern edge of North Andes block (five earth-quakes Mb > 6:0 from 1964 to 1995). The transferof deformation inland and the NE movement of theN. Andean block appear to be consequences of in-creased coupling from the Carnegie Ridge collision.In view of this significant activity, the deformation atthe southern limit of the North Andes block shouldbe quantified and the seismic risk along these faultsreassessed.

Acknowledgements

M.-A. Gutscher’s post-doctoral research wasfunded by a European Union TMR (Training andMobility of Researchers) Stipend. Many thanks toJ.-P. Eissen and E. Bourdon for stimulating discus-sions. Figures 2, 3, 4, 6 were produced using Wesseland Smith’s GMT software. [AC]

References

[1] J.L. Swenson, S.L. Beck, Historical 1942 Ecuador and 1942Peru subduction earthquakes, and earthquake cycles alongColombia–Ecuador and Peru subduction segments, PureAppl. Geophys. 146 (1996) 67–101.

[2] S.L. Beck, L.J. Ruff, The rupture process of the great 1979Colombia earthquake: evidence for the asperity model, J.Geophys. Res. 89 (1984) 9281–9291.

[3] W.R. McCann, S.P. Nishenko, L.R. Sykes, J. Krause, Seis-mic gaps and plate tectonics: seismic potential for majorboundaries, Pure Appl. Geophys. 117 (1979) 1083–1147.

[4] L.J. Ruff, Large earthquakes in subduction zones; segmentinteraction and recurrence times, in: G.E. Bebout, D.W.Scholl, S.H. Kirby, J.P. Platt (Eds.), Subduction: Top toBottom, American Geophysical Union, Washington, DC,1996, pp. 91–104.

[5] M. Cloos, R.L. Shreve, Shear-zone thickness and the seis-micity of Chilean and Marianas-type subduction zones,Geology 24 (1996) 107–110.

[6] C.H. Scholz, C. Small, The effect of seamount subductionon seismic coupling, Geology 25 (1997) 487–490.

[7] J.N. Kellogg, V. Vega, Tectonic development of Panama,Costa Rica, and the Colombian Andes: constraints fromGlobal Positioning System geodetic studies and gravity, in:


P. Mann (Ed.), Geologic and Tectonic Development of theCaribbean Plate Boundary in Southern Central America,Geol. Soc. Am. Spec. Pap. 295 (1995) 75–90.

[8] C. DeMets, R.G. Gordon, D.F. Angus, C. Stein, Currentplate motions , Geophys. J. Int. 101 (1990) 425–478.

[9] J.T. Freymueller, J.N. Kellogg, V. Vega, Plate motions in theNorth Andean region, J. Geophys. Res. 98 (1993) 21853–21863.

[10] T. De Vries, The geology of late Cenozoic marine terraces(tablazos) in northwestern Peru, J. S. Am. Earth Sci. 1(1988) 121–136.

[11] J.-F. Dumont, S. Benitez, et al., Neotectonics of the coastalregion of Ecuador: a new pluridisciplinary research project,in: P.R. Cobbold et al. (Eds.), Third International Sympo-sium on Andean Geodynamics (ISAG), St. Malo, Sept. 96,ORSTOM, France, 1996, pp. 175–178.

[12] R.N. Hey, Tectonic evolution of the Cocos–Nazca spreadingcenter, Geol. Soc. Am. Bull. 88 (1977) 1414–1420.

[13] P. Lonsdale, K.D. Klitgord, Structure and tectonic history ofthe eastern Panama Basin, Geol. Soc. Am. Bull. 89 (1978)981–999.

[14] T.W. Gardner, D. Verdonck, N.M. Pinter, R. Slingerland,K.P. Furlong, T.F. Bullard, S.G. Wells, Quaternary upliftastride the aseismic Cocos Ridge, Pacific coast, Costa Rica,Geol. Soc. Am. Bull. 104 (1992) 219–232.

[15] W.D. Pennington, Subduction of the eastern Panama Basinand seismotectonics of northwestern South America, J.Geophys. Res. 86 (1981) 10753–10770.

[16] R.A. Kolarsky, P. Mann, W. Montero, Island arc responseto shallow subduction of the Cocos Ridge, Costa Rica, in:P. Mann (Ed.), Geologic and Tectonic Development of theCaribbean Plate Boundary in Southern Central America,Geol. Soc. Am. Spec. Pap. 295 (1995) 235–262.

[17] N.C. Hardy, Tectonic evolution of the easternmost PanamaBasin: some new data and inferences, J. S. Am. Earth Sci.4 (1991) 261–269.

[18] E.R. Engdahl, R.D. van der Hilst, R. Buland, Global tele-seismic earthquake relocation with improved travel timesand procedures for depth relocation, Bull. Seismol. Soc.Am. 88 (1998) 722–743. Online Data Set (availablefrom E. Robert Engdahl, USGS) via anonymous ftp: ftp136.177.20.1 cd misc/engdahl/EHB compressed hypocenterdata file 1964–1995 (EHB.HDF) and format description(FORMAT.HDF)

[19] B.L. Askew, S.T. Algermissen (Eds.), Catalog of Earth-quakes for South America: Hypocenter and Intensity Data.Ceresis Publication Volumes 4, 6 and 7a, b and c (1985).Online Data Set (Both SISRA and PDE Catalogs avail-able from USGS National Earthquake Information Center):http://wwwneic.cr.usgs.gov/neis/epic/epic.html

[20] M.L. Hall, C.A. Wood, Volcano-tectonic segmentation ofthe N. Andes, Geology 13 (1985) 203–207.

[21] R. van der Hilst, P. Mann, Tectonic implications of tomo-graphic images of subducted lithosphere beneath northwest-ern South America, Geology 22 (1994) 451–454.

[22] R. Prevot, J.-L. Chatelain, B. Guillier, H. Yepes, Tomogra-phie des Andes equatoriennes: evidence d’une continuite

des Andes Centrales, C.R. Acad. Sci. Paris 323 (1996)833–840.

[23] M. Barazangi, B. Isacks, Spatial distribution of earthquakesand subduction of the Nazca plate beneath South America,Geology 4 (1976) 686–692.

[24] T. Cahill, B.L. Isacks, Seismicity and shape of the sub-ducted Nazca Plate, J. Geophys. Res. 97 (1992) 17503–17529.

[25] E.R. Engdahl, R.D. van der Hilst, J. Berrocal, Imagingof subducted lithosphere beneath South America, Geophys.Res. Lett. 22 (1995) 2317–2320.

[26] R.H. Pilger, Plate reconstructions, aseismic ridges, andlow-angle subduction beneath the Andes, Geol. Soc. Am.Bull. 92 (1981) 448–456.

[27] M.-A. Gutscher, J.-L. Olivet, D. Aslanian, J.-P. Eissen, R.Maury, Flat subduction beneath Peru caused by lost oceanicplateau, Nature (1999) in press.

[28] G. Suarez, P. Molnar, B.C. Burchfiel, Seismicity, fault planesolutions, depth of faulting, and active tectonics of theAndes of Peru, Ecuador and S. Colombia, J. Geophys. Res.88 (1983) 10403–10428.

[29] B. Parsons, J.G. Sclater, An analysis of the variation of theocean floor bathymetry and heat flow with age, J. Geophys.Res. 82 (1977) 803–827.

[30] A.M. Dziewonski, T.-A. Chou, J.H. Woodhouse, Deter-mination of earthquake source parameters from waveformdata for studies of global and regional seismicity, J. Geo-phys. Res. 86 (1981) 2825–2852. Online Data Set (avail-able from Harvard University Centroid Moment Tensor –CMT Project): http://www.seismology.harvard.edu/projects/CMT/CMTsearch.html

[31] G.L. Shepherd, R. Moberly, Coastal structure of the conti-nental margin, northwest Peru and southwest Ecuador, in:L. Kulm, J. Dymond, E.J. Dasch, D.M. Hussong (Eds.),Nazca Plate: Crustal Formation and Andean Convergence,Geol. Soc. Am. Mem. 154 (1981) 351–391.

[32] M. Daly, Correlations between Nazca=Farallon plate kine-matics and forearc basin evolution in Ecuador, Tectonics 8(1989) 769–790.

[33] A. Lavenu, T. Winter, F. Davilla, A Pliocene–Quaternarycompressional basin in the Interandean Depression, CentralEcuador, Geophys. J. Int. 121 (1995) 279–300.

[34] D.G. Herd, T.L. Youd, H. Meyer, C.J.L. Arango, W.J. Per-son, C. Mendoza, The Great Tumaco, Colombia Earthquakeof 12 December 1979, Science 211 (1981) 441–445.

[35] T.A. Minshull, S.C. Singh, G.K. Westbrook, Seismic ve-locity structure at a gas hydrate reflector, offshore westernColombia, from waveform inversion, J. Geophys. Res. 99(1994) 4715–4734.

[36] N.P. Mountney, G.K. Westbrook, Quantitative analysis ofMiocene to Recent forearc basin evolution along theColombian convergent margin, Basin Res. 9 (1997) 177–196.

[37] S. McGeary, A. Nur, Z. Ben-Avraham, Spatial gaps in arcvolcanism: the effect of collision or subduction of oceanicplateaus, Tectonophysics 119 (1985) 195–221.

[38] J. Vanek, V. Vankova, V. Hanus, Geochemical zonation of


volcanic rocks and deep structure of Ecuador and southernColombia, J. S. Am. Earth Sci. 7 (1994) 57–67.

[39] N.T. Arndt, A.C. Kerr, J. Tarney, Dynamic melting in plumeheads; the formation of Gorgona komatiites and basalts,Earth Planet. Sci. Lett. 146 (1997) 289–301.

[40] C.W. Sinton, R.A. Duncan, M. Storey, J. Lewis, J.J.Estrada, An oceanic flood basalt province within the Carib-bean plate, Earth Planet. Sci. Lett. 155 (1998) 221–235.

[41] R.H. Pilger, Cenozoic plate kinematics, subduction andmagmatism: South American Andes, J. Geol. Soc. London141 (1984) 793–802.

[42] M. Monzier, C. Robin, M.L. Hall, J. Cotten, P. Mothes,J.-P. Eissen, P. Samaniego, Les adakites d’Equateur: mod-ele preliminaire, C.R. Acad. Sci. Paris 324 (1997) 545–552.

[43] R. Barragan, D. Geist, M. Hall, P. Larson, M. Kurz,Subduction controls on the composition of lavas from theEcuadorian Andes, Earth Planet. Sci. Lett. 154 (1998)153–166.

[44] S.T. Johnston, D.J. Thorkelson, Cocos–Nazca slab windowbeneath Central America, Earth Planet. Sci. Lett. 146(1997) 465–474.

[45] M. Cloos, Lithospheric buoyancy and collisional orogen-

esis: subduction of oceanic plateaus, continental margins,island arcs, spreading ridges, and seamounts, Geol. Soc.Am. Bull. 105 (1993) 715–737.

[46] M.A. Feighner, M.A. Richards, Lithospheric structure andcompensation mechanisms of the Galapagos Archipelago,J. Geophys. Res. 99 (1994) 6711–6729.

[47] S. Nishenko, Circum-Pacific seismic potential: 1989–1999,Pure Appl. Geophys. 135 (1991) 169–259.

[48] F. Ego, M. Sebrier, A. Lavenu, H. Yepes, A. Eques,Quaternary state of stress in the northern Andes andthe restraining bend model for the Ecuadorian Andes,Tectonophysics 259 (1996) 101–116.

[49] T. Winter, J.-P. Avouac, A. Lavenu, Late Quaternary kine-matics of the Pallatanga strike-slip fault (Central Ecuador)from topographic measurements of displaced morphologicalfeatures, Geophys. J. Int. 115 (1993) 905–920.

[50] Smithsonian Global Volcanism Program, Online Data Set(available from Smithsonian Institution): http://www.volcano.si.edu/gvp/volcdata/index.htm

[51] W.H.F. Smith, D.T. Sandwell, Global seafloor topographyfrom satellite altimetry and ship depth soundings, Science277 (1997) 1956–1962.

Documents

Tectonic segmentation of the North Andean margin: impact of the